Chilkat Examples

ChilkatHOMEAndroid™Classic ASPCC++C#Mono C#.NET Core C#C# UWP/WinRTDataFlexDelphi ActiveXDelphi DLLVisual FoxProJavaLianjaMFCObjective-CPerlPHP ActiveXPHP ExtensionPowerBuilderPowerShellPureBasicCkPythonChilkat2-PythonRubySQL ServerSwift 2Swift 3/4TclUnicode CUnicode C++Visual Basic 6.0VB.NETVB.NET UWP/WinRTVBScriptXojo PluginNode.jsExcelGo

MFC Examples

Web API Categories

ASN.1
Amazon Glacier
Amazon S3
Amazon S3 (new)
Amazon SES
Amazon SNS
Amazon SQS
Async
Azure Cloud Storage
Azure Service Bus
Base64
Bounced Email
Box
CAdES
CSR
CSV
Certificates
Compression
DKIM / DomainKey
DSA
Diffie-Hellman
Digital Signatures
Dropbox
Dynamics CRM
ECC
Email Object
Encryption
FTP
FileAccess
Firebase
GMail REST API
Geolocation
Google APIs
Google Calendar
Google Cloud Storage
Google Drive
Google Photos
Google Sheets
Google Tasks
Gzip
HTML-to-XML/Text

HTTP
HTTP Misc
IMAP
JSON
JSON Web Encryption (JWE)
JSON Web Signatures (JWS)
JSON Web Token (JWT)
Java KeyStore (JKS)
MHT / HTML Email
MIME
Microsoft Graph
NTLM
OAuth1
OAuth2
OneDrive
OpenSSL
Outlook
PEM
PFX/P12
POP3
PRNG
REST
REST Misc
RSA Encryption
SCP
SFTP
SMTP
SSH
SSH Key
SSH Tunnel
SharePoint
Socket/SSL/TLS
Spider
Stream
Tar Archive
Upload
WebSocket
XAdES
XML
XML Digital Signatures
XMP
Zip
curl

 

 

 

(MFC) HTML Table to CSV

Demonstrates a method for converting an HTML table to a CSV file.

Note: This example requires Chilkat v9.5.0.77 or greater.

Chilkat C/C++ Library Downloads

MS Visual C/C++ Libs

See Also: Using MFC CString in Chilkat

#include <CkHttp.h>
#include <CkBinData.h>
#include <CkHtmlToXml.h>
#include <CkStringBuilder.h>
#include <CkXml.h>
#include <CkCsv.h>

void ChilkatSample(void)
    {
    CkString strOut;

    // This example requires the Chilkat API to have been previously unlocked.
    // See Global Unlock Sample for sample code.

    // First download the HTML containing the table
    CkHttp http;
    CkBinData bdHtml;

    bool success = http.QuickGetBd("https://example-code.com/data/etf_table.html",bdHtml);
    if (success != true) {
        strOut.append(http.lastErrorText());
        strOut.append("\r\n");
        SetDlgItemText(IDC_EDIT1,strOut.getUnicode());
        return;
    }

    // Convert to XML.
    CkHtmlToXml htx;
    htx.SetHtmlBd(bdHtml);

    CkStringBuilder sbXml;
    htx.ToXmlSb(sbXml);

    CkXml xml;
    xml.LoadSb(sbXml,true);

    // Remove attributes and sub-trees we don't need.
    // (In other words, we're getting rid of clutter...)
    int numRemoved = xml.PruneTag("thead");
    numRemoved = xml.PruneAttribute("style");
    numRemoved = xml.PruneAttribute("class");

    // Scrub the element and attribute content.
    xml.Scrub("ContentTrimEnds,ContentTrimInside,AttrTrimEnds,AttrTrimInside");

    // Let's see what we have...
    strOut.append(xml.getXml());
    strOut.append("\r\n");

    // We have the following XML.
    // Copy this XML into the online tool at Generate Parsing Code from XML
    // as a starting point for accessing the data..

    // <?xml version="1.0" encoding="utf-8"?>
    // <root>
    //     <html>
    //         <head>
    //             <meta http-equiv="content-type" content="text/html; charset=UTF-8"/>
    //         </head>
    //         <body text="#000000" bgcolor="#FFFFFF">
    //             <div>
    //                 <div>
    //                     <table role="grid" data-scrollx="true" data-sortdirection="desc" data-sorton="-1"/>
    //                 </div>
    //             </div>
    //             <div>
    //                 <table id="topHoldingsTable" role="grid" data-scrollx="true" data-sortdirection="desc" data-sorton="-1">
    //                     <tbody>
    //                         <tr role="row">
    //                             <td>
    //                                 <text>ITUB4</text>
    //                             </td>
    //                             <td>
    //                                 <text>ITAU UNIBANCO HOLDING PREF SA</text>
    //                             </td>
    //                             <td>
    //                                 <text>Financials</text>
    //                             </td>
    //                             <td>
    //                                 <text>Brazil</text>
    //                             </td>
    //                             <td>
    //                                 <text>10.94</text>
    //                             </td>
    //                             <td>
    //                                 <text>998,954,813.73</text>
    //                             </td>
    //                         </tr>
    //                         <tr role="row">
    //                             <td>
    //                                 <text>BBDC4</text>
    //                             </td>
    //                             <td>
    //                                 <text>BANCO BRADESCO PREF SA</text>
    //                             </td>
    //                             <td>
    //                                 <text>Financials</text>
    //                             </td>
    //                             <td>
    //                                 <text>Brazil</text>
    //                             </td>
    //                             <td>
    //                                 <text>9.01</text>
    //                             </td>
    //                             <td>
    //                                 <text>822,164,622.75</text>
    //                             </td>
    //                         </tr>
    // 			...
    // 			...
    // 			...
    //                     </tbody>
    //                 </table>
    //             </div>
    //         </body>
    //     </html>
    // </root>

    // 
    // This is the code generated by the online tool:
    // 
    int i;
    int count_i;
    const char *table_role = 0;
    const char *table_data_scrollx = 0;
    const char *table_data_sortdirection = 0;
    const char *table_data_sorton = 0;
    const char *table_id = 0;
    int j;
    int count_j;
    const char *tr_role = 0;
    int k;
    int count_k;
    const char *tagPath = 0;
    const char *text = 0;

    i = 0;
    count_i = xml.NumChildrenHavingTag("html|body|div");
    while (i < count_i) {
        xml.put_I(i);
        table_role = xml.chilkatPath("html|body|div[i]|div|table|(role)");
        table_data_scrollx = xml.chilkatPath("html|body|div[i]|div|table|(data-scrollx)");
        table_data_sortdirection = xml.chilkatPath("html|body|div[i]|div|table|(data-sortdirection)");
        table_data_sorton = xml.chilkatPath("html|body|div[i]|div|table|(data-sorton)");
        table_id = xml.chilkatPath("html|body|div[i]|table|(id)");
        table_role = xml.chilkatPath("html|body|div[i]|table|(role)");
        table_data_scrollx = xml.chilkatPath("html|body|div[i]|table|(data-scrollx)");
        table_data_sortdirection = xml.chilkatPath("html|body|div[i]|table|(data-sortdirection)");
        table_data_sorton = xml.chilkatPath("html|body|div[i]|table|(data-sorton)");
        j = 0;
        count_j = xml.NumChildrenHavingTag("html|body|div[i]|table|tbody|tr");
        while (j < count_j) {
            xml.put_J(j);
            tr_role = xml.chilkatPath("html|body|div[i]|table|tbody|tr[j]|(role)");
            k = 0;
            count_k = xml.NumChildrenHavingTag("html|body|div[i]|table|tbody|tr[j]|td");
            while (k < count_k) {
                xml.put_K(k);
                text = xml.getChildContent("html|body|div[i]|table|tbody|tr[j]|td[k]|text");
                k = k + 1;
            }

            j = j + 1;
        }

        i = i + 1;
    }

    // Let's modify the above code to build the CSV.
    CkCsv csv;
    csv.SetColumnName(0,"Ticker");
    csv.SetColumnName(1,"Name");
    csv.SetColumnName(2,"Sector");
    csv.SetColumnName(3,"Country");
    csv.SetColumnName(4,"Weight");
    csv.SetColumnName(5,"Notional Vaue");

    i = 0;
    count_i = xml.NumChildrenHavingTag("html|body|div");
    while (i < count_i) {
        xml.put_I(i);
        j = 0;
        count_j = xml.NumChildrenHavingTag("html|body|div[i]|table|tbody|tr");
        while (j < count_j) {
            xml.put_J(j);
            k = 0;
            count_k = xml.NumChildrenHavingTag("html|body|div[i]|table|tbody|tr[j]|td");
            while (k < count_k) {
                xml.put_K(k);
                csv.SetCell(j,k,xml.getChildContent("html|body|div[i]|table|tbody|tr[j]|td[k]|text"));
                k = k + 1;
            }

            j = j + 1;
        }

        i = i + 1;
    }

    csv.SaveFile("qa_output/brasil_etf.csv");
    const char *csvStr = csv.saveToString();
    strOut.append(csvStr);
    strOut.append("\r\n");

    // Our CSV looks like this:
    // Ticker,Name,Sector,Country,Weight,Notional Vaue
    // ITUB4,ITAU UNIBANCO HOLDING PREF SA,Financials,Brazil,10.94,"998,954,813.73"
    // BBDC4,BANCO BRADESCO PREF SA,Financials,Brazil,9.01,"822,164,622.75"
    // VALE3,CIA VALE DO RIO DOCE SH,Materials,Brazil,8.60,"785,290,260.07"
    // PETR4,PETROLEO BRASILEIRO PREF SA,Energy,Brazil,5.68,"518,124,434.10"
    // PETR3,PETROBRAS,Energy,Brazil,4.86,"443,254,438.53"
    // B3SA3,B3 BRASIL BOLSA BALCAO SA,Financials,Brazil,4.57,"417,636,740.16"
    // ABEV3,AMBEV SA,Consumer Staples,Brazil,4.57,"417,216,913.63"
    // BBAS3,BANCO DO BRASIL SA,Financials,Brazil,3.25,"296,921,232.15"
    // ITSA4,ITAUSA INVESTIMENTOS ITAU PREF SA,Financials,Brazil,2.90,"265,153,684.52"
    // LREN3,LOJAS RENNER SA,Consumer Discretionary,Brazil,2.25,"205,832,175.98"
    // 


    SetDlgItemText(IDC_EDIT1,strOut.getUnicode());

    }

 

© 2000-2019 Chilkat Software, Inc. All Rights Reserved.